Slight improvement over v2.0, using different training params. No changes in prompt compared to previous version. Dataset is exactly the same too.
SDXL version of my Liselotte lora, trained on pony diffusion and animagineXL 3.0
